Please note that is project is still ongoing, all project files including this documentation are being updated constantly at the time being.
- Project Overview
- Obtaining Demo Code
- Testing Data (Optional)
- Running the Demo Locally using Docker
- Running the Demo in your local workspace
- Verbose Results
The purpose of this project was to experiment with streamlit and docker for an image stitching experiment using OpenCV's newly added Stitcher class, which is based on methods proposed in David G. Lowe's paper.
To get started clone the demo code to your local device and navigate to that directory.
git clone https://github.com/Mo-637/image-stitching.git
cd image-stitching
Optional: Testing Data (Source)
Since the app relies on a GUI to select the input images to stitch you'd be free to use your own image sequences and samples. In case data is not readily you may opt to downloading Adobe's Panorama Dataset. This dataset contains high quality sequences of ten outdoor scenes.
Download and extract the test data it into the data folder within the cloned work directory.
chmod +x data/data_download.sh
./data/data_download.sh
To follow though with this demo, please make sure that you have Git and Docker installed beforehand.
NOTE: If you're using a Linux machine please make sure you also follow Docker's post-installation steps
here
or
here
to be able to run docker commands without needing sudo priveleges.
- Build the docker image
docker build -t 'stitcher:latest' .
- As a sanity check, make sure the image has been built and is listed using
docker images ls
- Run a container with the built image
docker container run -p 8501:8501 -d stitcher:latest
- Access the demo by visiting the following address in your preferred browser
http://172.17.0.2:8501/
- To stop the running container(s)
docker ps -a -q
If you prefer running the files in your workspace simply
- Install the prerequisites (Streamlit, OpenCV 3/4 and Numpy)
python3 -m pip install requirements.txt
- Run the Demo
streamlit run demo/gui.py